A power distribution network power quality comprehensive treatment system based on harmonic tracing

Abstract

The present application relates to the technical field of power quality treatment, and particularly relates to a power quality comprehensive treatment system of distribution network based on harmonic tracing, comprising: a transient recording wave module, which is arranged at a user property boundary point and a public connection point, and is used for triggering bidirectional synchronous recording wave when a harmonic mutation variable exceeds a limit, generating a transient waveform with an absolute time scale and extracting a time-frequency domain transient fingerprint feature; a precursor sensing module, which is arranged at an impact load body, and is used for sensing a heterogeneous physical precursor signal before harmonic current establishment; and an edge computing module, which is used for determining a responsible subject according to a time sequence and a similarity of the transient fingerprint feature and generating a harmonic responsibility evidence. In the present application, the non-intrusive right confirmation of a harmonic impact event is realized, and the problem that an existing steady-state responsibility allocation method depends on accurate line parameters and impedance models and is prone to misjudging background harmonics as user emission when parameters are mismatched, thereby leading to incorrect treatment and reward and punishment disputes is solved.

Description

Technical Field

[0001] This invention relates to the field of power quality management technology, and in particular to a comprehensive power quality management system for distribution networks based on harmonic source tracing. Background Technology

[0002] Power quality in a distribution network refers to the degree to which the waveforms, frequencies, and amplitudes of voltage and current deviate from ideal sine waves and nominal values. Among these, harmonic distortion is one of the main power quality issues affecting power supply reliability and equipment safety. With the widespread integration of nonlinear loads such as distributed renewable energy sources, electric arc furnaces, and variable frequency drives into distribution networks, harmonic pollution is showing a trend towards more complex sources, more fluctuating impacts, and a wider range of influence, placing higher demands on the comprehensive management of power quality in distribution networks.

[0003] In existing technologies, power quality management systems for distribution networks typically employ harmonic state estimation or point of common coupling responsibility allocation methods. Synchronous monitoring units are deployed at key nodes, and the harmonic contribution of each feeder or user is calculated by constructing a harmonic admittance matrix or analyzing fluctuations. Based on this, active power filters and other devices are dispatched for centralized or local compensation to reduce the harmonic voltage distortion rate of the entire network.

[0004] However, the following problems still exist in the existing technology: I. Existing methods for assigning harmonic responsibility heavily rely on accurate line impedance parameters and operating topology. When distribution network parameters shift due to temperature, aging, and random switching of distributed power sources, the steady-state decomposition results can easily misjudge system background harmonics as user-side emissions, leading to the responsibility determination conclusions not being accepted by users and the inability to implement harmonic reward and punishment mechanisms.

[0005] Second, existing active filter devices adopt a closed-loop feedback control mode, which requires the detection of harmonic current before the calculation of compensation command. The delay generated by the entire acquisition, communication and processing link causes the compensation current output to lag behind the rapid fluctuation of the impact load. Under dynamic operating conditions, the compensation effect is significantly reduced. In severe cases, the compensation current and the real-time harmonics are in phase and superimposed, which amplifies the harmonics.

[0006] Therefore, to solve the above problems, a comprehensive power quality management system for distribution networks based on harmonic source tracing is proposed. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides the following technical solution: A comprehensive power quality management system for distribution networks based on harmonic source tracing includes: The transient waveform recording module is deployed at the user property boundary point and the common connection point. It is used to trigger bidirectional synchronous waveform recording when the harmonic mutation exceeds the limit, generate transient waveforms with absolute time scales and extract transient fingerprint features in the time and frequency domain.

[0008] The precursor sensing module is deployed on the impulsive load body to sense heterogeneous physical precursor signals before the establishment of harmonic current.

[0009] The edge computing module is used to determine the responsible party and generate harmonic responsibility evidence based on the temporal sequence and similarity of the transient fingerprint features, and at the same time identify the load condition type from the heterogeneous physical precursor signals and generate feedforward instructions.

[0010] The governance knowledge base module, deployed on a cloud platform, is used to generate a library of compensation current templates corresponding to the operating conditions when the load is in steady state, and to slowly iterate and update the templates using feedback from the compensated residual current.

[0011] The distributed active filtering module is used to receive the feedforward instruction and call the corresponding template to perform advance compensation in an open-loop feedforward manner.

[0012] The slow iterative updates of the governance knowledge base module and the open-loop feedforward compensation of the distributed active filter module constitute a decoupled control system of fast feedforward and slow feedback.

[0013] As an improvement to the above technical solution: The edge computing module determines the responsible party based on transient fingerprint features. Specifically, it compares the timescales of the transient fingerprint features extracted by the transient recording module at the property boundary point and the common connection point. When the time difference between the two is within a preset reasonable range and the similarity of the fingerprint features meets the preset conditions, the user on the corresponding line is determined to be the responsible party for this harmonic impact event.

[0014] As an improvement to the above technical solution: The transient fingerprint feature extracted by the transient recording module is a multi-dimensional feature vector that reflects the harmonic energy distribution and phase jump characteristics after time-frequency transformation of the waveforms of several cycles before and after the sudden change. The feature vector is used to distinguish harmonic impact events generated by different nonlinear devices.

[0015] As an improvement to the above technical solution: The harmonic liability evidence is used to generate non-repudiation electronic evidence, which includes transient waveform segments, extracted feature data, and judgment conclusions, and supports two-way verification by the user end and the power grid end.

[0016] As an improvement to the above technical solution: The precursor sensing module is used to acquire the non-electrical quantity signal intrinsically generated by the impact load during the production process before the harmonic current increases significantly. The timing difference between the non-electrical quantity signal and the harmonic current is used to buy time for subsequent compensation preparation.

[0017] As an improvement to the above technical solution: The feedforward instruction generated by the edge computing module includes target harmonic order and compensation phase interval information, and is sent to the distributed active filter module through a deterministic communication network.

[0018] As an improvement to the above technical solution: The open-loop feedforward compensation of the distributed active filter module and the slow closed-loop iterative update of the governance knowledge base module operate on a decoupled control time scale.

[0019] The open-loop feedforward compensation utilizes the lead time of the precursor signal to complete the preparation, and the slow closed-loop iterative update is performed with a relatively long period.

[0020] As an improvement to the above technical solution: The governance knowledge base module is also used to extract transient fingerprint features and operating condition types corresponding to the impact loads identified as responsible entities based on the harmonic responsibility evidence generated by the edge computing module, and generate personalized compensation strategies for the responsible entities.

[0021] When the early warning sensing module identifies the same type of working condition belonging to the same responsible party again, the distributed active filtering module calls the personalized compensation strategy to perform compensation.

[0022] The beneficial effects of this invention are: 1. In this invention, transient recording modules are deployed at user property boundary points and common connection points to extract transient fingerprint feature vectors. An edge computing module then compares the causal timing and cosine similarity of the fingerprint timescales on both sides. This achieves non-intrusive determination of responsibility for harmonic impact events, solving the problem that existing steady-state responsibility allocation methods rely on precise line parameters and impedance models, and are prone to misjudging background harmonics as user emissions when parameters are mismatched, leading to erroneous governance and disputes over rewards and punishments.

[0023] 2. In this invention, a precursor sensing module is deployed on the impact load to capture heterogeneous physical precursor signals before the establishment of harmonic current. An edge computing module generates feedforward instructions, driving a distributed active filter module to perform advance compensation in an open-loop feedforward manner. Simultaneously, a governance knowledge base module uses the residual current feedback after compensation to perform template iterative updates at a slow cycle. This achieves timing matching between the compensation response and harmonic changes, as well as decoupling control between fast feedforward and slow feedback. It solves the problem in existing closed-loop feedback governance schemes where the compensation current lags behind dynamic harmonic changes due to detection, calculation, and communication delays, and even amplifies harmonics due to phase misalignment superposition. Attached Figure Description

[0024] Figure 1 This is a system architecture diagram of the present invention. Detailed Implementation

[0025] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0026] See appendix Figure 1 A comprehensive power quality management system for distribution networks based on harmonic source tracing, comprising: The transient waveform recording module is deployed at the user property boundary point and the common connection point. It is used to trigger bidirectional synchronous waveform recording when the harmonic mutation exceeds the limit, generate transient waveforms with absolute time scales and extract transient fingerprint features in the time and frequency domain.

[0027] The precursor sensing module is deployed on the impulsive load body to sense heterogeneous physical precursor signals before the establishment of harmonic current.

[0028] The edge computing module is used to determine the responsible party based on the timing and similarity of transient fingerprint features and generate harmonic responsibility evidence. At the same time, it identifies the load condition type from heterogeneous physical precursor signals and generates feedforward instructions.

[0029] The governance knowledge base module, deployed on a cloud platform, is used to generate a library of compensation current templates corresponding to the operating conditions when the load is in steady state, and to slowly iterate and update the templates using feedback from the compensated residual current.

[0030] The distributed active filtering module is used to receive feedforward instructions and call the corresponding template to perform advance compensation in an open-loop feedforward manner.

[0031] The slow iterative updates of the governance knowledge base module and the open-loop feedforward compensation of the distributed active filter module constitute a decoupled control system of fast feedforward and slow feedback.

[0032] In one embodiment, when an electric arc furnace user in the power distribution network is in the melting period, the electrodes frequently come into contact with and break off from the scrap steel, generating severe harmonic impacts.

[0033] The transient waveform recording module deployed at the user's incoming line cabinet and the transient waveform recording module at the bus common connection point simultaneously detected the fifth harmonic current mutation exceeding the limit, and immediately triggered bidirectional synchronous waveform recording, each generating transient waveform data with GPS absolute time stamp.

[0034] Meanwhile, the vibration sensor installed on the electrode lifting mechanism of the electric arc furnace captures the characteristic mechanical vibration signal the moment the electrode contacts the scrap steel. The appearance of this signal precedes the significant establishment of the harmonic current by about 100 milliseconds.

[0035] After receiving the transient fingerprint features from both sides, the edge computing module compares their time scales and finds that the fingerprint on the user side leads the fingerprint on the bus side. The time difference matches the propagation delay of the cable segment, and the multidimensional similarity of the fingerprints on both sides reaches 97%. Therefore, it determines that the electric arc furnace user is the responsible party for this harmonic impact and generates an unalterable harmonic responsibility certificate.

[0036] Meanwhile, the edge computing module identifies the current material collapse arcing condition from the spectral characteristics of the vibration signal, generates a trigger frame containing the 5th and 7th harmonic compensation targets and corresponding phase intervals, and sends it to the distributed active filter module at the bus via industrial Ethernet. In the next control cycle after receiving the trigger frame, the filter module calls the compensation current template corresponding to the material collapse arcing condition from the stored compensation current template library and outputs the reverse harmonic current in an open-loop feedforward manner.

[0037] After the compensation is completed, the filtering module uploads the residual current data to the governance knowledge base module of the cloud platform. This module fine-tunes and optimizes the amplitude and phase of the compensation current template at minute-level intervals. The updated template is called in the next similar operating condition, forming a closed-loop evolution of fast feedforward instantaneous response and slow feedback gradually approaching the optimal.

[0038] See appendix Figure 1 The edge computing module determines the responsible party based on transient fingerprint features. Specifically, it compares the timescales of the transient fingerprint features extracted by the transient recording module at the property boundary point and the common connection point. When the time difference between the two is within a preset reasonable range and the similarity of the fingerprint features meets the preset conditions, the user on the corresponding line is determined to be the responsible party for this harmonic impact event.

[0039] In one embodiment, when the total distortion rate of the bus voltage suddenly exceeds the standard, a transient waveform recording module deployed at the property boundary point of a steel rolling workshop and a transient waveform recording module deployed at the common connection point simultaneously record a waveform containing impulse characteristics.

[0040] The edge computing module first extracts the timestamps of the fingerprint features from both sides and calculates the measured time difference. The wavelength is 65 microseconds. An investigation revealed that the cable length is approximately 120 meters, and the electromagnetic wave propagation speed is... Taking 0.8 times the speed of light, calculate the theoretical propagation delay. Approximately 0.5 microseconds. Substitute the parameters into the formula for the decision criteria:

[0041] The obtained deviation value is within the allowable deviation threshold after considering waveform recording trigger jitter. Within this range, the causal time sequence condition holds.

[0042] Subsequently, the edge computing module performed cosine similarity calculation on the fingerprint features on both sides, and the result was 0.96, which exceeded the preset threshold of 0.90. Based on the causal time sequence, it was determined that the steel rolling workshop was the sole responsible party for this harmonic impact event, and the determination result was packaged together with the waveform recordings on both sides for evidence storage.

[0043] See appendix Figure 1 The transient fingerprint feature extracted by the transient waveform recording module is a multi-dimensional feature vector that reflects the harmonic energy distribution and phase jump characteristics after time-frequency transformation of the waveform of several cycles before and after the sudden change. The feature vector is used to distinguish harmonic impact events generated by different nonlinear devices.

[0044] In one embodiment, for the harmonic impact generated at the moment of lifting by a variable frequency crane, the transient waveform recording module extracts the waveforms of 5 cycles before and after the abrupt change point and performs time-frequency transformation. The transient fingerprint feature vector F_u is extracted from the user side, and the transient fingerprint feature vector F_p is extracted from the common connection point side. Both contain three components: the peak value of the 5th harmonic energy, the phase jump variable of the 7th harmonic, and the duration of the interharmonic energy cluster.

[0045] The edge computing module calculates the cosine similarity S between F_u and F_p, and the result is 0.96, which is greater than the preset similarity threshold S_th of 0.90. Therefore, it is determined that the fingerprint features on both sides belong to the same harmonic impact event.

[0046] The multidimensional feature vector is highly matched with the fingerprint of the variable frequency crane start-up already labeled in the database, while it is significantly different from the fingerprint features of the electric arc furnace on the same bus, thus realizing the accurate identification of the impact source equipment.

[0047] See appendix Figure 1 Harmonic liability evidence storage is used to generate non-repudiable electronic evidence. The electronic evidence includes transient waveform fragments, extracted feature data, and judgment conclusions, and supports two-way verification between the user end and the power grid end.

[0048] In one embodiment, for example: a chemical company's rectifier device generates an occasional harmonic impact due to an abnormal trigger pulse. After the edge computing module determines that the company is the responsible party, it automatically generates a report containing 10 waveform segments before and after the impact time, the property boundary point, and the common connection point, 10 cycles of waveform recordings from two synchronous channels, the extracted 5th and 7th harmonic phase transition characteristic values, and a causal determination logic explanation.

[0049] The report's hash value is uploaded to a blockchain platform, allowing companies to access the stored report through their own backend systems and independently verify the waveform segments and feature data using publicly available verification tools. This confirms the objectivity of the judgment conclusions without relying on unilateral monitoring data from the power grid.

[0050] See appendix Figure 1The precursor sensing module is used to acquire the non-electrical quantity signal generated by the impact load during the production process before the harmonic current increases significantly. The timing difference between the non-electrical quantity signal and the harmonic current is used to buy time for subsequent compensation preparation.

[0051] In one embodiment, during the scrap steel melting stage of electric arc furnace smelting, when the electrode descends and contacts the scrap steel, it first generates mechanical collision vibration and sharp friction sound. After about 80 to 150 milliseconds, the electric arc stabilizes and generates strong harmonic current that is injected into the power grid.

[0052] An accelerometer installed on the electrode lifting hydraulic cylinder immediately outputs a trigger signal when the vibration signal exceeds a set threshold. This signal is transmitted via hardwire to the edge computing module, which identifies the event of the electrode contacting scrap steel from the vibration signal.

[0053] By taking advantage of this inherent timing difference of tens of milliseconds, the edge computing module has ample time to complete the operation condition identification and instruction generation, enabling the distributed active filter module to complete the output preparation of the compensation current before the harmonic current impact arrives.

[0054] See appendix Figure 1 The feedforward instructions generated by the edge computing module contain information on the target harmonic order and the compensation phase interval, and are sent to the distributed active filter module through a deterministic communication network.

[0055] In one embodiment, the lightweight neural network within the edge computing module identifies an impending collapse and arc ignition condition from the characteristic sound of scrap steel collapsing in the furnace captured by the sound intensity probe array, and then generates a feedforward instruction. The data structure of this instruction includes the following fields: the target harmonic order is 5th and 7th, the 5th order compensation phase interval is 120 to 135 degrees lagging, and the 7th order compensation phase interval is 30 to 45 degrees leading.

[0056] The instruction frame is forwarded with the highest priority through an industrial Ethernet switch that supports time-sensitive networking. After passing through two network devices, it reaches the communication interface of the distributed active filter module, with an end-to-end transmission delay of no more than 500 microseconds.

[0057] After parsing the instruction, the filter module immediately indexes the compensation current template for the corresponding operating condition from memory and loads it into the inverter output buffer.

[0058] See appendix Figure 1 The open-loop feedforward compensation and governance knowledge base module of the distributed active filter module operate on a decoupled control time scale.

[0059] Open-loop feedforward compensation utilizes the lead time of precursor signals to complete preparation, while slow closed-loop iterative updates are performed over a relatively long period.

[0060] In one embodiment, when the precursor sensing module detects the refining furnace electrode lifting signal, the distributed active filter module starts to output the pre-stored refining end compensation current template in the second control cycle after receiving the feedforward instruction. The entire open-loop feedforward process takes about 3 milliseconds from the triggering of the precursor signal to the output of the compensation current, which completely covers the rising edge of the subsequent harmonic current.

[0061] Meanwhile, the governance knowledge base module deployed on the cloud platform continuously receives residual current waveform data after compensation uploaded by the filtering module via the 4G network in the background. Every 60 seconds, a batch processing optimization algorithm is run. Based on the residual statistics of all compensation events within that time period, the compensation current template parameters at the end of the refining process are iteratively updated. The update rules are as follows:

[0062] In the formula, For the first The updated compensation current template parameters. For the first The parameters after the last update For the first The residual current error signal after compensation, μ is the convergence step size coefficient.

[0063] When the refining furnace enters the same operating conditions again, the filter module will use an updated template, and the compensation accuracy will gradually improve over time.

[0064] See appendix Figure 1 The governance knowledge base module is also used to extract transient fingerprint features and operating condition types corresponding to the impact loads identified as responsible entities based on the harmonic responsibility evidence generated by the edge computing module, and generate personalized compensation strategies for the responsible entities.

[0065] When the early warning module identifies the same type of operating condition belonging to the same responsible party again, the distributed active filter module calls the personalized compensation strategy to perform compensation.

[0066] In one embodiment, after summarizing harmonic liability evidence data for a consecutive week, the governance knowledge base module discovered through cluster analysis that the high-order harmonics generated by the roughing mill of a large steel rolling mill at the moment of steel biting have a unique time-frequency fingerprint, and the impact event has been repeatedly determined to be the responsibility of the steel rolling mill.

[0067] The governance knowledge base module then marked the steel biting condition of the roughing mill as a separate responsibility condition category, and combined it with the best-performing compensation current templates under this condition from historical compensation data to generate a unique personalized compensation strategy library for the steel mill, which was then distributed to the distributed active filter module of the corresponding bus.

[0068] Subsequently, whenever the vibration sensor installed on the roughing mill stand captures the characteristic vibration signal of steel biting, the precursor sensing module identifies it as a marked responsible working condition. The distributed active filtering module directly calls the compensation current template in the personalized compensation strategy library corresponding to the steel mill to perform compensation, realizing closed-loop management of responsible entity identification and compensation strategy customization.

[0069] The above embodiments are merely illustrative of the technical solutions of the present invention and are not intended to limit it. Anyone skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A comprehensive power quality management system for distribution networks based on harmonic source tracing, characterized in that, include: The transient waveform recording module is deployed at the user property boundary point and the common connection point. It is used to trigger bidirectional synchronous waveform recording when the harmonic mutation exceeds the limit, generate transient waveforms with absolute time scales and extract transient fingerprint features in the time and frequency domain. The precursor sensing module is deployed on the impulsive load body to sense heterogeneous physical precursor signals before the establishment of harmonic current; The edge computing module is used to determine the responsible party and generate harmonic responsibility evidence based on the temporal sequence and similarity of the transient fingerprint features, and at the same time identify the load condition type from the heterogeneous physical precursor signal and generate feedforward instructions. The governance knowledge base module, deployed on the cloud platform, is used to generate a compensation current template library corresponding to the operating conditions when the load is in steady state, and to slowly iterate and update the templates using the feedback of the compensated residual current. A distributed active filter module is used to receive the feedforward instruction and call the corresponding template to perform advance compensation in an open-loop feedforward manner. The slow iterative updates of the governance knowledge base module and the open-loop feedforward compensation of the distributed active filter module constitute a decoupled control system of fast feedforward and slow feedback.

2. The comprehensive power quality management system for distribution networks based on harmonic source tracing according to claim 1, characterized in that: The edge computing module determines the responsible party based on transient fingerprint features. Specifically, it compares the timescales of the transient fingerprint features extracted by the transient recording module at the property boundary point and the common connection point. When the time difference between the two is within a preset reasonable range and the similarity of the fingerprint features meets the preset conditions, the user on the corresponding line is determined to be the responsible party for this harmonic impact event.

3. The comprehensive power quality management system for distribution networks based on harmonic source tracing as described in claim 1, characterized in that: The transient fingerprint feature extracted by the transient recording module is a multi-dimensional feature vector that reflects the harmonic energy distribution and phase jump characteristics after time-frequency transformation of the waveforms of several cycles before and after the sudden change. The feature vector is used to distinguish harmonic impact events generated by different nonlinear devices.

4. The comprehensive power quality management system for distribution networks based on harmonic source tracing according to claim 1, characterized in that: The harmonic liability evidence is used to generate non-repudiation electronic evidence, which includes transient waveform segments, extracted feature data, and judgment conclusions, and supports two-way verification by the user end and the power grid end.

5. A comprehensive power quality management system for distribution networks based on harmonic source tracing as described in claim 1, characterized in that: The precursor sensing module is used to acquire the non-electrical quantity signal intrinsically generated by the impact load during the production process before the harmonic current increases significantly. The timing difference between the non-electrical quantity signal and the harmonic current is used to buy time for subsequent compensation preparation.

6. A comprehensive power quality management system for distribution networks based on harmonic source tracing as described in claim 1, characterized in that: The feedforward instruction generated by the edge computing module includes target harmonic order and compensation phase interval information, and is sent to the distributed active filter module through a deterministic communication network.

7. A comprehensive power quality management system for distribution networks based on harmonic source tracing as described in claim 1, characterized in that: The open-loop feedforward compensation of the distributed active filter module and the slow closed-loop iterative update of the governance knowledge base module operate on a decoupled control time scale. The open-loop feedforward compensation utilizes the lead time of the precursor signal to complete the preparation, and the slow closed-loop iterative update is performed with a relatively long period.

8. A comprehensive power quality management system for distribution networks based on harmonic source tracing as described in claim 1, characterized in that: The governance knowledge base module is also used to extract transient fingerprint features and operating condition types corresponding to the impact loads identified as responsible entities based on the harmonic responsibility evidence generated by the edge computing module, and generate personalized compensation strategies for the responsible entities. When the early warning sensing module identifies the same type of working condition belonging to the same responsible party again, the distributed active filtering module calls the personalized compensation strategy to perform compensation.

Images